[Topic-models] Intuition behind CTM and DTM

David Mimno mimno at cs.umass.edu
Mon Nov 24 15:13:54 EST 2008


Thanks Erik! 

Note to reviewers: I don't use those functions for research purposes. :)

On Mon, Nov 24, 2008 at 11:56:03AM -0800, Erik Sudderth wrote:
> Thanks for the very clear explanation of the logistic normal 
> distribution, David.  I noticed a small bug in your R code: you need to 
> multiply by a square root of the covariance matrix, not the matrix 
> itself, to sample from a Gaussian.  The following code does this:


More information about the Topic-models mailing list